Adaptive image database using wavelets

We propose a novel approach to image database design, using wavelets and their inherent hierarchical nature. Simple feature vectors are computed from the images and used to tune the database storage structure, yielding the maximum performance. This approach is more efficient than conventional approaches in terms of retrieval rates and storage space. The database needs less number of parameters to be stored and less computation has to be done to retrieve images which are similar to a query image. Moreover, the simplicity of the work leads to a design of a more scalable system.